There can be few fields of human endeavour in which history counts for so little as in the world of finance.
Numbers sometimes seem emphatic and inflexible, especially when used to convey complex issues of general interest. They can give a spurious impression of certainty even when they reflect subjective judgements or approximations. Risk estimations are a case in point. Risk ratings may be based on anything from a totally subjective view expressed through a simple risk matrix, through to the complexity of Neural Network methodologies. To make effective use of this variably reliable information, recipients need to be aware of its quality.
Misunderstanding about risk isn’t helped by the fact that there is no transcending standard unit of measurement for it. The cover design on books about risk, which draw heavily on tumbling dice, spinning roulette wheels and chess pieces, give an impression of clear reasoning, simple probabilities and reliably scalable prediction. In fact, risk measurement in the financial world and in Health and Safety has none of those qualities.
There are no agreed standard units of risk; no measurement that can be applied equally to the risk of dangerous roads, the risk in different diets, the risk of air travel, the risk of an investment or epidemic. Risk across its wide spectrum is chaotically quantified in pragmatic and idiosyncratic ways; in terms that reflect the situation and context:
This illustrates the reality that the concept of RISK lacks overall coherence; rather there are many distinct ‘pockets’ of coherence, each with its own focus, methods, measurement and terminology. Insurance, book-making, road safety, health, investment, diet, aeronautics or seismology, for example.
Risk predictions are typically expressed as probabilities, but these are never akin to the ‘classic’ or ‘a priori’ probability, as in the throw of a dice (so ignore the book covers). A priori predictions are true and certain because all the possible outcomes are known and each is equally likely. Their distinctive feature, as every maths pupil knows, is that no matter how many times the penny comes up ‘heads’ the probability of getting the same result yet again is still 50:50. Counting incidents, i.e. what has happened in the past, has no bearing at all on an a priori probability. Yet counting incidents and events from the past is exactly the basis for the probabilities used in predicting all real world risk. This kind of probability is not the same thing at all. For obvious reasons, this can only ever provide an approximate and relatively unreliable estimate. Statistical models and Neural Network methodologies are tools that support the most sophisticated predictions available. Their creators will be highly skilled, very intelligent and able to incorporate wisdom, experience and expert judgement into the writing of their algorithms. Nevertheless, the raw material is always and of necessity, historic data. This situation has been characterised as attempting to drive a car while looking only in the rear view mirror. It’s possible, so long as the road ahead looks like the road you’ve passed, and so long as you draw the right inferences about what is likely to happen. But you’re not going to see what’s coming if it’s something entirely different.
Risk professionals in the driving seat hopefully understand these limitations of risk estimates and remain alert to the completely unexpected. The problem for the rest of us is that numbers carry a lot of weight and authority. Since we deal with such critical matters relatively rarely, and tend to rely on the advice of others, it is our own decisions about finance and safety that are vulnerable to the apparently emphatic risk statements of experts and statisticians.
All of the above focuses on the nature of risk; but what about the nature of those creating the risk or being exposed to the risk? This has increasingly been recognised as a very significant issue, not least by the financial regulators. In addition to the requirement to make the risk involved in any investment or financial product clear to their clients, intermediaries are now also required to take the client’s risk appetite into account. There is an important role here for Risk Type, both to clarify what investments would be appropriate, but also to help the client to better appreciate the implications of their own risk dispositions; how much risk would they be comfortable with?… how would they react if returns fall short of expectation?… how resilient would they be to the ups and downs of the market?… will their highly optimistic outlook get them into trouble?… will their haste and impatience with detail prevent proper scrutiny?… will their anxieties interfere with good decision making? How do they compare to others in these respects? These are all personality issues and these things are all knowable.
Perhaps surprisingly, the human side of the risk equation turns out to be very coherent; significantly more so than the tangled conglomerate that is RISK per se. After many thousands of years of success and survival, homo sapiens has ironed out these fundamentally crucial aspects of human nature. Nature’s answer has been to provide our species with a rich variety of risk dispositions that are complementary to one another, building the formidable ‘Team Homo-Sapiens’. Our species is equipped, in equal measure, with people that are adventurous, carefree, excitable, intense, wary, prudent, deliberate and composed. Every one of them has an important contribution to make. These are the characteristics that shape the driving, investing, road crossing, sporting, purchasing, entrepreneurial, voting behaviour that accounts for the statistics, that feed the algorithms on which risk model building relies.
Maybe it’s a good idea to give greater priority to self-knowledge and insight into the diverse risk disposition of others before trying to master all the problematic complexities and uncertainties of the risk itself?
Geoff Trickey, September 2018